Highly interconnected enhancer communities control lineage-determining genes in human mesenchymal stem cells

Abstract

Adipocyte differentiation is driven by waves of transcriptional regulators that reprogram the enhancer landscape and change the wiring of the promoter interactome. Here, we use high-throughput chromosome conformation enhancer capture to interrogate the role of enhancer-to-enhancer interactions during differentiation of human mesenchymal stem cells. We find that enhancers form an elaborate network that is dynamic during differentiation and coupled with changes in enhancer activity. Transcription factors (TFs) at baited enhancers amplify TF binding at target enhancers, a phenomenon we term cross-interaction stabilization of TFs. Moreover, highly interconnected enhancers (HICE) act as integration hubs orchestrating differentiation by the formation of three-dimensional enhancer communities, inside which, HICE, and other enhancers, converge on phenotypically important gene promoters. Collectively, these results indicate that enhancer interactions play a key role in the regulation of enhancer function, and that HICE are important for both signal integration and compartmentalization of the genome.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Enhancer-capture Hi-C identifies functional chromatin interactions in hMSCs.
Fig. 2: The enhancer interactome is highly plastic and linked to enhancer activity at both ends of an interaction.
Fig. 3: Transcription factor cross-talk between connected enhancers.
Fig. 4: HICE engage in multiple types of chromatin interactions.
Fig. 5: HICE communities predict dynamic gene expression during adipogenesis.
Fig. 6: Regulation of HICE communities defines lineage choice.

Data availability

All generated sequence data are available at the GEO repository under accession code GSE140782. In addition to the data generated for this study, raw sequencing data from single-cell and bulk RNA-seq, ChIP–seq and DNase I-seq were downloaded from the GEO under accession code GSE113253.

References

  1. 1.

    Siersbaek, R. et al. Transcription factor cooperativity in early adipogenic hotspots and super-enhancers. Cell Rep. 7, 1443–1455 (2014).

    CAS  Google Scholar 

  2. 2.

    Waki, H. et al. Global mapping of cell-type-specific open chromatin by FAIRE-seq reveals the regulatory role of the NFI family in adipocyte differentiation. PLoS Genet. 7, e1002311 (2011).

    CAS  Google Scholar 

  3. 3.

    Steger, D. J. et al. Propagation of adipogenic signals through an epigenomic transition state. Genes Dev. 24, 1035–1044 (2010).

    CAS  Google Scholar 

  4. 4.

    Mikkelsen, T. S. et al. Comparative epigenomic analysis of murine and human adipogenesis. Cell 143, 156–169 (2010).

    CAS  Google Scholar 

  5. 5.

    Siersbaek, R. et al. Extensive chromatin remodelling and establishment of transcription factor ‘hotspots’ during early adipogenesis. EMBO J. 30, 1459–1472 (2011).

    CAS  Google Scholar 

  6. 6.

    Rauch, A. et al. Osteogenesis depends on commissioning of a network of stem cell transcription factors that act as repressors of adipogenesis. Nat. Genet. 51, 716–727 (2019).

  7. 7.

    Siersbaek, R. et al. Dynamic rewiring of promoter-anchored chromatin loops during adipocyte differentiation. Mol. Cell 66, 420–435 (2017).

    CAS  Google Scholar 

  8. 8.

    Freire-Pritchett, P. et al. Global reorganisation of cis-regulatory units upon lineage commitment of human embryonic stem cells. eLife 6, e21926 (2017).

    Google Scholar 

  9. 9.

    Rubin, A. J. et al. Lineage-specific dynamic and preestablished enhancer-promoter contacts cooperate in terminal differentiation. Nat. Genet. 49, 1522–1528 (2017).

    CAS  Google Scholar 

  10. 10.

    de Laat, W. & Duboule, D. Topology of mammalian developmental enhancers and their regulatory landscapes. Nature 502, 499–506 (2013).

    Google Scholar 

  11. 11.

    Degner, J. F. et al. DNase I sensitivity QTLs are a major determinant of human expression variation. Nature 482, 390–394 (2012).

    CAS  Google Scholar 

  12. 12.

    Grubert, F. et al. Genetic control of chromatin states in humans involves local and distal chromosomal interactions. Cell 162, 1051–1065 (2015).

    CAS  Google Scholar 

  13. 13.

    Huang, J. et al. Dissecting super-enhancer hierarchy based on chromatin interactions. Nat. Commun. 9, 943 (2018).

    Google Scholar 

  14. 14.

    Markenscoff-Papadimitriou, E. et al. Enhancer interaction networks as a means for singular olfactory receptor expression. Cell 159, 543–557 (2014).

    CAS  Google Scholar 

  15. 15.

    Ing-Simmons, E. et al. Spatial enhancer clustering and regulation of enhancer-proximal genes by cohesin. Genome Res. 25, 504–513 (2015).

    CAS  Google Scholar 

  16. 16.

    Pliner, H. A. et al. Cicero predicts cis-regulatory DNA interactions from single-cell chromatin accessibility data. Mol. Cell 71, 858–871 (2018).

    CAS  Google Scholar 

  17. 17.

    Gate, R. E. et al. Genetic determinants of co-accessible chromatin regions in activated T cells across humans. Nat. Genet. 50, 1140–1150 (2018).

    CAS  Google Scholar 

  18. 18.

    Carithers, L. J. et al. A novel approach to high-quality postmortem tissue procurement: the GTEx project. Biopreserv. Biobank. 13, 311–319 (2015).

    Google Scholar 

  19. 19.

    Zhang, W. et al. The TEA domain family transcription factor TEAD4 represses murine adipogenesis by recruiting the cofactors VGLL4 and CtBP2 into a transcriptional complex. J. Biol. Chem. 293, 17119–17134 (2018).

    CAS  Google Scholar 

  20. 20.

    Otto, F. et al. Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell 89, 765–771 (1997).

    CAS  Google Scholar 

  21. 21.

    Tanaka, T., Yoshida, N., Kishimoto, T. & Akira, S. Defective adipocyte differentiation in mice lacking the C/EBPβ and/or C/EBPδ gene. EMBO J. 16, 7432–7443 (1997).

    CAS  Google Scholar 

  22. 22.

    Gubelmann, C. et al. Identification of the transcription factor ZEB1 as a central component of the adipogenic gene regulatory network. eLife 3, e03346 (2014).

    Google Scholar 

  23. 23.

    Dowen, J. M. et al. Control of cell identity genes occurs in insulated neighborhoods in mammalian chromosomes. Cell 159, 374–387 (2014).

    CAS  Google Scholar 

  24. 24.

    Zuin, J. et al. Cohesin and CTCF differentially affect chromatin architecture and gene expression in human cells. Proc. Natl Acad. Sci. USA 111, 996–1001 (2014).

    CAS  Google Scholar 

  25. 25.

    Hnisz, D. et al. Super-enhancers in the control of cell identity and disease. Cell 155, 934–947 (2013).

    CAS  Google Scholar 

  26. 26.

    Whyte, W. A. et al. Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell 153, 307–319 (2013).

    CAS  Google Scholar 

  27. 27.

    Schoenfelder, S. et al. The pluripotent regulatory circuitry connecting promoters to their long-range interacting elements. Genome Res. 25, 582–597 (2015).

    CAS  Google Scholar 

  28. 28.

    Javierre, B. M. et al. Lineage-specific genome architecture links enhancers and non-coding disease variants to target gene promoters. Cell 167, 1369–1384 (2016).

    CAS  Google Scholar 

  29. 29.

    Joshi, O. et al. Dynamic reorganization of extremely long-range promoter–promoter interactions between two states of pluripotency. Cell Stem Cell 17, 748–757 (2015).

    CAS  Google Scholar 

  30. 30.

    Perry, M. W., Boettiger, A. N. & Levine, M. Multiple enhancers ensure precision of gap gene-expression patterns in the Drosophila embryo. Proc. Natl Acad. Sci. USA 108, 13570–13575 (2011).

    CAS  Google Scholar 

  31. 31.

    Zhou, Q. et al. Onset of atonal expression in Drosophila retinal progenitors involves redundant and synergistic contributions of Ey/Pax6 and So binding sites within two distant enhancers. Dev. Biol. 386, 152–164 (2014).

    CAS  Google Scholar 

  32. 32.

    Stine, Z. E., McGaughey, D. M., Bessling, S. L., Li, S. & McCallion, A. S. Steroid hormone modulation of RET through two estrogen responsive enhancers in breast cancer. Hum. Mol. Genet. 20, 3746–3756 (2011).

    CAS  Google Scholar 

  33. 33.

    Huang, J. et al. Dynamic control of enhancer repertoires drives lineage and stage-specific transcription during hematopoiesis. Dev. Cell. 36, 9–23 (2016).

    Google Scholar 

  34. 34.

    Oudelaar, A. M. et al. Single-allele chromatin interactions identify regulatory hubs in dynamic compartmentalized domains. Nat. Genet. 50, 1744–1751 (2018).

    CAS  Google Scholar 

  35. 35.

    Allahyar, A. et al. Enhancer hubs and loop collisions identified from single-allele topologies. Nat. Genet. 50, 1151–1160 (2018).

    CAS  Google Scholar 

  36. 36.

    Huang, J., Marco, E., Pinello, L. & Yuan, G.-C. Predicting chromatin organization using histone marks. Genome Biol. 16, 162 (2015).

    Google Scholar 

  37. 37.

    Schmitt, A. D. et al. A compendium of chromatin contact maps reveals spatially active regions in the human genome. Cell Rep. 17, 2042–2059 (2016).

    CAS  Google Scholar 

  38. 38.

    Li, T., Jia, L., Cao, Y., Chen, Q. & Li, C. OCEAN-C: mapping hubs of open chromatin interactions across the genome reveals gene regulatory networks. Genome Biol. 19, 54 (2018).

    Google Scholar 

  39. 39.

    Alipour, E. & Marko, J. F. Self-organization of domain structures by DNA-loop-extruding enzymes. Nucleic Acids Res. 40, 11202–11212 (2012).

    CAS  Google Scholar 

  40. 40.

    Fudenberg, G. et al. Formation of chromosomal domains by loop extrusion. Cell Rep. 15, 2038–2049 (2016).

    CAS  Google Scholar 

  41. 41.

    Hnisz, D., Shrinivas, K., Young, R. A., Chakraborty, A. K. & Sharp, P. A. A phase separation model for transcriptional control. Cell 169, 13–23 (2017).

    CAS  Google Scholar 

  42. 42.

    Sabari, B. R. et al. Co-activator condensation at super-enhancers links phase separation and gene control. Science 361, eaar3958 (2018).

  43. 43.

    Gong, Y. et al. Stratification of TAD boundaries reveals preferential insulation of super-enhancers by strong boundaries. Nat. Commun. 9, 542 (2018).

    Google Scholar 

  44. 44.

    Schwarzer, W. et al. Two independent modes of chromatin organization revealed by cohesin removal. Nature 551, 51–56 (2017).

    Google Scholar 

  45. 45.

    Bintu, B. et al. Super-resolution chromatin tracing reveals domains and cooperative interactions in single cells. Science 362, eaau1783 (2018).

  46. 46.

    Nuebler, J., Fudenberg, G., Imakaev, M., Abdennur, N. & Mirny, L. A. Chromatin organization by an interplay of loop extrusion and compartmental segregation. Proc. Natl Acad. Sci. USA 115, E6697–E6706 (2018).

    CAS  Google Scholar 

  47. 47.

    Simonsen, J. L. et al. Telomerase expression extends the proliferative life-span and maintains the osteogenic potential of human bone marrow stromal cells. Nat. Biotechnol. 20, 592–596 (2002).

    CAS  Google Scholar 

  48. 48.

    Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).

    CAS  Google Scholar 

  49. 49.

    Madsen, J. et al. iRNA-seq: computational method for genome-wide assessment of acute transcriptional regulation from total RNA-seq data. Nucleic Acids Res. 43, e40 (2015).

  50. 50.

    McCarthy, D. J., Chen, Y. & Smyth, G. K. Differential expression analysis of multifactor RNA-seq experiments with respect to biological variation. Nucleic Acids Res. 40, 4288–4297 (2012).

    CAS  Google Scholar 

  51. 51.

    Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies and species. Nat. Biotechnol. 36, 411–420 (2018).

    CAS  Google Scholar 

  52. 52.

    Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 32, 381–386 (2014).

    CAS  Google Scholar 

  53. 53.

    Kuhn, M. Building predictive models in R using the caret package. J. Stat. Softw. 28, 1–26 (2008).

    Google Scholar 

  54. 54.

    Zhang, Y. et al. Model-based analysis of ChIP-seq (MACS). Genome Biol. 9, R137 (2008).

    Google Scholar 

  55. 55.

    Li, Q., Brown, J. B., Huang, H. & Bickel, P. J. Measuring reproducibility of high-throughput experiments. Ann. Appl. Stat. 5, 1752–1779 (2011).

    Google Scholar 

  56. 56.

    Heinz, S. et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589 (2010).

    CAS  Google Scholar 

  57. 57.

    Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923–930 (2014).

    CAS  Google Scholar 

  58. 58.

    Skene, P. J. & Henikoff, S. An efficient targeted nuclease strategy for high-resolution mapping of DNA binding sites. eLife 6, e21856 (2017).

    Google Scholar 

  59. 59.

    Nagano, T. et al. Comparison of Hi-C results using in-solution versus in-nucleus ligation. Genome Biol. 16, 175 (2015).

    Google Scholar 

  60. 60.

    Wingett, S. et al. HiCUP: pipeline for mapping and processing Hi-C data. F1000Res. 4, 1310 (2015).

    Google Scholar 

  61. 61.

    Wang, X. runHiC: a user-friendly Hi-C data processing software based on hiclib (2016); https://doi.org/10.5281/zenodo.55324

  62. 62.

    Wang, X.-T., Cui, W. & Peng, C. HiTAD: detecting the structural and functional hierarchies of topologically associating domains from chromatin interactions. Nucleic Acids Res. 45, e163 (2017).

    Google Scholar 

  63. 63.

    Cairns, J. et al. CHiCAGO: robust detection of DNA looping interactions in Capture Hi-C data. Genome Biol. 17, 127 (2016).

    Google Scholar 

  64. 64.

    Csardi, G. & Nepusz, T. The igraph software package for complex network research. InterJournal Complex Systems, 1695 (2006).

Download references

Acknowledgements

This work was supported by grants from the Independent Research Fund Denmark (Sapere Aude Advanced grant no. 12-125524), the Danish National Research Foundation (DNRF grant no. 141) to the Center for Functional Genomics and Tissue Plasticity (ATLAS), the Novo Nordisk Foundation (Advanced Grant) and through grants to the Danish Diabetes Academy and NNF Center for Stem Cell Biology (NNF17CC0027852), the Villum Foundation (through support to the Villum Center for Bioanalytical Sciences) and the UKRI Medical Research Council (MR/L007150/1). We thank S. Andrews at the Babraham Institute for assistance with the initial probe design and our colleagues from the Functional Genomics and Metabolism Research Unit at the University of Southern Denmark for comments and discussions that greatly improved the manuscript.

Author information

Affiliations

Authors

Contributions

Conceptualization: J.G.S.M., A.R., A.H.K., S.T. and S.M. Experimental work: M.S.M., A.R., S.T., E.L.V.H., B.M.J. Formal analysis, investigation and data curation: J.G.S.M. Visualization: J.G.S.M. and M.H. Writing: J.G.S.M. and S.M. Funding acquisition: S.M. Supervision: S.M. and P.F.

Corresponding author

Correspondence to Susanne Mandrup.

Ethics declarations

Competing interests

P.F. is a cofounder of Enhanc3D Genomics. The remaining authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Related to Figure 1.

a, Principal component analysis of all ECHi-C replicates and conditions (D0 = day 0, D1 = day 1, D10 = day 10). b, Stratum-adjusted correlation coefficients between all replicates and conditions (D0 = day 0, D1 = day 1, D10 = day 10). c, Fraction of all interactions detected at the indicated levels of significance at day 0 after combining the replicates (n0.01 = 398,606, n0.001 = 225,480, n0.0001 = 140,545 interactions) that are found in both individual replicates (at FDR ≤ 0.05). d, Histogram of the log2 absolute distances between significantly interacting fragments.

Extended Data Fig. 2 Related to Figure 2.

a, Virtual 4C plots of an induced (left) and a repressed (right) interaction. Line plots show log transformed loess smoothened normalized signal in a 200 kilobase window around the viewpoint. The black line represents the mean of the replicates, the shaded grey area shows the spread of the replicates and the dotted red line show the expected signal derived from CHiCAGO47. Heat maps show log transformed normalized signal for both replicates for each pooled fragment inside the window. Arcs show detected interactions (FDR ≤ 0.01). b, Enrichment of dynamic enhancer-to-enhancer interactions connecting enhancers where with no enhancers are co-regulated with the interaction, one of the enhancers are co-regulated or both of the enhancers are co-regulated relative to constant enhancer-to-enhancer interactions connecting enhancers where either neither enhancer is dynamically regulated (0, n = 70,181), one of the enhancer is dynamic (1, n = 42,441) or both of the enhancers are dynamic (2, n = 15,204). Dynamic regulation was defined as a significant change (FDR ≤ 0.05) in MED1 signal between day 0 and day 1 or 10. Co-regulation was defined as dynamic enhancers where MED1 changes in the same direction (gained or lost) as the interaction. c, Example of enhancers that interact in hMSC-TERT4 and for which an epigenomic QTL has been demonstrated in lymphoblastoid cells. Left panel: MED1 occupancy (normalized tag counts) and the interaction in undifferentiated hMSCs (day 0). Right panel: Effect of the genotype of the variant on the accessibility of the target site across lymphoblastoid cells11. (Ref = Homozygous for reference genotype, Het = Heterozygous, Alt = Homozygous for alternative genotype). (nRef = 32, nHet = 27, nAlt = 8).

Extended Data Fig. 3 Related to Figure 3.

a, Linear regression showing Loess smoothened occupancy (normalized tag counts) of the indicated factors measured using CUT&RUN at bound (≥ 4-fold over input, normalized tag count ≥ 25% quantile) baited enhancers as a function of occupancy at their target enhancers (within 200 kb). The black line shows the signal, and the shaded area the standard error. R2: Pearson’s correlation coefficient. α: The slope of the linear regression between occupancy (normalized tag counts) at either end using smoothened data. b, Boxplot showing the occupancy (normalized tag counts) for the indicated TFs at day 0 or day 1 and the absolute interaction distances for occupancy- and distance-matched baited enhancers (defined as in Fig. 3d).

Extended Data Fig. 4 Related to Figure 4.

a, Top 5 most enriched motifs in the top 1000 least connected regular enhancers versus the top 1000 most connected HICE at any time point during differentiation. b, Fraction of fragments from the indicated Hi-C libraries in different quality categories as defined by HiCUP62. c, Number of valid and captured read pairs after HiCUP quality filtering for each replicate of Hi-C in undifferentiated hMSCs (day 0). d, Boxplot indicating domain sizes of TADs and subTADs. Domains were detected as described in Fig. 4e. e, Example of a TAD and a subTAD that is delineated by an enhancer-to-enhancer interaction anchored in a HICE. DI = Directionality index. f, Fraction of baited enhancers strongly bound by CTCF (≥ 10 normalized tags, n = 1,147) in undifferentiated hMSCs (day 0). Baited enhancers were stratified by being HICE or regular as well as being boundary or non-boundary. Boundary enhancers were defined as enhancers harboring cross domain interactions (defined as in Fig. 4e). All other enhancers were defined as non-boundary. g, The number of CTCF sites with induced (green: FDR ≤ 0.05, logFC > 0, n = 2,028), unchanged (grey, n = 14,773) and repressed (red: FDR ≤ 0.05, logFC < 0 n = 1,577) CTCF occupancy for different groups of enhancers defined by the indicated change in number of detected enhancer-to-enhancer interactions between day 1 and day 0 of differentiation.

Extended Data Fig. 5 Related to Figure 5.

a, Boxplot indicating the number of enhancer-to-promoter interactions (E-P) for promoters in regular communities (defined as in Fig. 5a), or for promoters in HICE communities with an increasing number of interactions used as threshold for defining HICE. b, Boxplot indicating the number of enhancers per promoter in regular communities (defined as in Fig. 5a) or in communities defined by containing at least one HICE with an increasing number of interactions as threshold for defining HICE. c, Number of enhancer-to-promoter interactions (E-P) for enhancers in regular communities or in HICE communities (defined as in Fig. 5a). d, Overlap between HICE and baited super-enhancer constituents. Super-enhancers were identified with HOMER58 using the MED1 signal in DNase I hypersensitive sites at day 0, day 1 or day 10 of adipocyte differentiation. Constituents were defined as DHS sites overlapping with a super-enhancer identified at any time points. N denotes the number of enhancers. e, Enrichment of differentially expressed genes (maximum normalized tag count ≥ 1 and FDR ≤ 0.05, |log2FC| ≥ 1.5 at day 1 or 10 relative to undifferentiated hMSCs (day 0), n = 5,069) among genes contacted by both HICE communities and super-enhancer constituents (+HICE, +SE, n = 488), only HICE communities (+HICE, -SE, n = 346) or only super-enhancer constituents (-HICE, +SE, n = 1,160) relative to 1000 permutations of randomly selected genes. Data shown as mean ± SD. Dots show individual data points. f, Expression of markers of stem cells, preadipocytes or adipocytes in hMSC-TERT4 single cells6 ordered by pseudo-time calculated using Monocle54. g, Accuracy (Acc) and Kappa index of prediction of class labels (stem cell, preadipocytes or adipocytes based on pseudo-time) using a random forest model. The random forest model was trained on 80% of cells in each class and using all expressed genes. The accuracy and Kappa were evaluated on the remaining 20% of cells that was not used for training.

Extended Data Fig. 6 Related to Figure 6.

a, Fraction of fragments from the indicated Hi-C libraries in different quality categories as defined by HiCUP62. b, The number of valid and captured read pairs after HiCUP quality filtering for each in day 1 osteoblasts. c, The fraction of enhancer-to-enhancer (E-E, n = 109,333) and enhancer-to-promoter (E-P, n = 20,945) interactions that are significantly changed (FDR ≤ 0.05) between day 0 of differentiation and either day 1 in adipogenesis or day 1 in osteogenesis. d, The number of significant (FDR ≤ 1%) enhancer-to-enhancer and enhancer-to-promoter interactions in day 1 osteoblasts (Ob), undifferentiated hMSCs (D0) and day 1 adipocytes (Ad). e, Overlap of baits defined as HICE (baits with at least 8 enhancer-to-enhancer interactions) throughout early time points of adipogenesis and osteogenesis. N denotes the number of baits. f, Pie chart showing the number of enhancer communities (#promoters/community = 0) and gene regulatory communities (#promoters/community > 0). The bar plot shows the number of HICE gene regulatory communities (#HICE/community > 0) and regular gene regulatory communities (#HICE/community = 0). Communities were detected based on label propagation through a network constructed from all enhancers and promoters (nodes) and interactions (edges) weighted by their significance level. g, The log2 enrichment of enhancer-to-enhancer interactions located in the indicated clusters within HICE communities enriched for enhancer-to-promoter interactions from either cluster 2, 5 or 6 (as defined in Fig. 6b).

Supplementary information

Supplementary Information

Supplementary Notes 1–6 and Figs. 1–4

Reporting Summary

Supplementary Tables

Supplementary Table 1

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Madsen, J.G.S., Madsen, M.S., Rauch, A. et al. Highly interconnected enhancer communities control lineage-determining genes in human mesenchymal stem cells. Nat Genet 52, 1227–1238 (2020). https://doi.org/10.1038/s41588-020-0709-z

Download citation

Search

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing